Pattern mining from movement of mobile users

Abstract

With the rapid progress of mobile technology, an increasing number of mobile devices are able to be tracked. In the field of cellular technology, we can track the location of a mobile phone user by locating the cell connected with his/her mobile phone. When a mobile user moves from one location to another, we can track his/her mobile device connecting from one cell to another cell. This work focuses on mining patterns from mobile user movement data. Two new algorithms are proposed, namely: location link and user link pattern mining algorithms. Both proposed algorithms are able to mine a pattern from another pattern instead of mining a pattern directly from a data source. Therefore, the pattern will be more concise and the processing time of both algorithms will be faster. Both algorithms have been evaluated from two perspectives: pattern mining generation and time processing. The experiment results from the processing time aspect indicate that larger datasets and lower values of user-defined thresholds lead to a longer processing time for both algorithms. Furthermore, the processing time of both proposed algorithms are dominated by the preparation process.